Speech

App works great. Parkeeter TDT V3 not so much. I speak Polish and it detects Russian. Also there is no possibility to force language: > unfortuntately parakeet-v3 model doesn;t recieve or output language id https://github.com/NVIDIA-NeMo/Speech/issues/14799#issuecomm... Also this: > If you are using Parakeet for English only then you should be using V2. V3 is for several languages and is worse at English only. https://news.ycombinator.com/item?id=48897908

You only need the ASR bits -- this is where I got to when I previously looked into running Parakeet: # NeMo does not run on 3.13+ python3.12 -m venv .venv source .venv/bin/activate git clone https://github.com/NVIDIA/NeMo.git nemo cd nemo pip install torch torchaudio torchvision --index-url https://download.pytorch.org/whl/cu128 pip install .[asr] deactivate Then run a transcribe.py script in that venv: import os import sys import nemo.collections.asr as nemo_asr model_path = sys.argv[1] audio_path = sys.argv[2] # Load from a local path... asr_model = nemo_asr.models.EncDecRNNTBPEModel.restore_from(restore_path=model_path) # Or download from huggingface ('org/model')... asr_model = nemo_asr.models.EncDecRNNTBPEModel.from_pretrained(model_name=model_path) outpu

With how hard something like this is, we should be looking at the positive side. Picovoice here solves both the issues of hotword/wake-word detection and intent extraction. This looks like something you could build on top of ARM's [keyword spotting program](https://github.com/ARM-software/ML-KWS-for-MCU) and the wake word services listed in [Rhasspy's docs](https://rhasspy.readthedocs.io/en/latest/wake-word/#raven) But, to implement something like this from scratch would take a good while. Also, here are some Automatic Speech Recognition toolkits (which won't run offline on a microcontroller) out there. These are useful to pipe the data into a program that deals with intents (something like [RASA](https://rasa.com) (Require Internet) * [Deepgram](https://deepgram.com) - I believe they build upon OpenAI's Whisper model and have their own c

I think there's been talk to do speaker diarization with whisper-asr-webservice[0] which is also written in python and should be able to make use of goodies such as pyannote-audio, py-webrtcvad, etc. Whisper is great but at the point we get to kludging various things together it might start to make more sense to use something like Nvidia NeMo[1] which was built with all of this in mind and more. [0] - https://github.com/ahmetoner/whisper-asr-webservice [1] - https://github.com/NVIDIA/NeMo

Out of interest, did you try Nemo? https://github.com/NVIDIA/NeMo

I worked with this a bit not that long ago. For cloud services, quality of Google and Azure "neural" voices are tough to beat. Interestingly I experienced significant latency for all of the Azure services regardless of region, configuration, etc. Never dug deep enough to figure out what was going on there. Also of note, Azure will let you run their implementation on a local container with the usual "contact us" stuff. Not sure of the terms and pricing on that. For local, Mozilla TTS was best from a quality standpoint but the GPU inference support was a bit dicey and (possibly) not really supported at all. For more complex and bespoke applications the Nvidia (I know, I know) NeMO toolkit [0] is very powerful but requires more effort than most to get up and running. However, it provides the ability to do very interesting things with additio

https://github.com/NVIDIA/NeMo which is open source, Pytorch based and regularly publishes new models and checkpoints.

NVidia NeMo: https://github.com/NVIDIA/NeMo

plenty open-source options are available. Checkout https://github.com/NVIDIA/NeMo

Interesting project! There is a similar concept/implementation of Neural Types in NVIDIA NeMo https://github.com/NVIDIA/NeMo/blob/main/tests/core/test_typ... that you might find interesting

Have a look at NeMo https://github.com/nvidia/NeMo it comes with QuartzNet (only 19M of weights and better accuracy than DeepSpeech2) pretrained on thousands of hours of speech.

If you want type safety on top of Pytorch, checkout this project https://github.com/NVIDIA/NeMo

NVIDIA has QuartzNet which contains only 19M weights and achieves 3.9% on test clean without language model and less then 3 with LM. Code (Pytorch): https://github.com/NVIDIA/NeMo Paper: https://arxiv.org/pdf/1910.10261.pdf